How Decoy Effect Pricing Changes Choice on Pricing Pages
Key takeaway
A decoy can change which plan people choose without increasing value per visitor. Keep comparisons clear and evaluate the randomized groups on contribution, refunds, and retention rather than target-plan share alone.
A pricing test can move more customers into your premium plan while making the business worse off. That happens when the higher premium share comes from losing basic-plan buyers, discounting customers who would have paid more, or attracting people who cannot use what they bought. A convincing pricing comparison and a profitable pricing comparison are separate things to establish.
The decoy effect gives us a hypothesis about the first. The experiment has to answer the second. My VALUE framework connects those questions: validate the target plan, analyze the value gap, limit the decoy strategically, understand customer segments, and experiment with positioning.
What the decoy effect actually predicts
Joel Huber, John Payne, and Christopher Puto’s original research on asymmetrically dominated alternatives found that adding an option could increase preference for the existing option that dominated it. The added alternative was worse than one choice, without being worse than every choice in the set.
That definition is narrower than “put three plans on a pricing page.” A middle plan can be a reasonable compromise. An expensive plan can provide a price anchor. Neither automatically demonstrates asymmetric dominance.
Consider this deliberately simplified illustration, not a recommended product lineup:
Assuming every other relevant attribute is identical, Pro dominates the comparison option: it provides more projects for the same price. Basic still trades fewer projects for a lower price. The comparison may make Pro easier to evaluate against Basic.
If the comparison option instead costs $119, Pro requires an extra $10. There is now a price-versus-capacity tradeoff. Calling that a pure decoy effect would overstate what the pricing structure establishes.
Real buyers also care about support, integrations, implementation work, and contract terms. Those attributes can undo a tidy two-column comparison. Use the research to formulate a testable explanation, rather than assume a familiar-looking pricing table will reproduce a laboratory result.
Step 1 — Validate your target plan performance
The V in VALUE asks whether the plan you want people to choose deserves more customers. Examine activation, actual feature use, refunds, retention, and the cost of serving buyers who already select it. Compare customers at similar ages and with similar needs; a mature enterprise cohort is a poor benchmark for newly acquired small teams.
There is no universal retention or gross-margin percentage that makes a plan suitable for a decoy test. The relevant threshold comes from your product, cost structure, and acquisition economics. A high-priced plan with expensive onboarding can produce less contribution than a simpler plan.
I would begin with a concrete question: which identifiable buyer benefits enough from this plan to justify its additional price? If the answer is only “the business needs more premium sales,” the pricing page is being asked to solve a product problem.
Set the business outcome before drawing the comparison table. For example, use contribution per eligible visitor over a stated observation window, with refund and cancellation guardrails. If contribution cannot yet be measured reliably, document the revenue proxy and the costs it omits.
Step 2 — Analyze the value gap between tiers
The A is about what the customer receives for the additional commitment. Translate feature counts into jobs: more projects for an agency managing several clients, permissions for a team with approval requirements, or an integration that removes recurring manual work.
Start with customer conversations and usage patterns, then check whether the distinctions are visible on the page. “Advanced analytics” does little comparison work if buyers cannot tell whether it solves their reporting problem. A clear description can make the existing lineup easier to understand without adding another plan.
Keep payment periods, usage limits, and add-on charges comparable. A lower headline price can conceal a higher total cost when essential features are charged separately. That is the practical connection to mental accounting and subscription bundling: how a cost is presented changes the comparison the buyer must perform.
Avoid assigning invented “value scores” to plans. If you use willingness-to-pay research, record the sample, question, and uncertainty. Otherwise, describe the tradeoff in terms customers can verify.
Step 3 — Limit the decoy strategically
The L means constrain the intervention, including the complexity it introduces. Do not create broken workflows, hidden restrictions, or misleading availability merely to make another plan look attractive. Every plan offered for purchase needs accurate terms and a product the business will deliver.
Get Atticus | Better Decisions’s stories in your inbox
Join Medium for free to get updates from this writer.
There is a useful distinction here. The simplified table above illustrates a research mechanism. A real commercial tier may have a legitimate purpose that makes it preferable for some customers. In that case, it may no longer be strictly dominated. Describe the experiment as a test of the comparison set rather than claim you have isolated a psychological mechanism.
Often the first worthwhile treatment is clearer positioning of existing plans. Keep the underlying prices and entitlements unchanged, make the additional benefit explicit, and test whether buyers understand it. The decision is not automatically improved by minimizing every word; a longer explanation can be appropriate when customers need it.
Before launch, check the full path from plan selection to invoice. A comparison that appears transparent on the pricing page can become misleading if checkout changes the period, quantity, or charge.
Step 4 — Understand your customer segments
The U addresses a common interpretation mistake: treating premium preference as evidence of better customer fit. A buyer with one small project may correctly prefer Basic even when Pro offers a lower cost per project. Unused capacity does not become valuable because the unit price looks attractive.
Define a small number of relevant segments before the experiment, such as new versus existing customers or individuals versus teams. Use characteristics known before exposure. Grouping people by the plan they selected afterward creates a selection effect: the treatment may itself change who enters each group.
Review the overall randomized comparison first. Segment results can help explain the pattern, but a positive subgroup discovered after examining many slices needs confirmation. A larger apparent premium shift on desktop might reflect audience differences, visibility, or chance. It does not by itself prove a device-specific mechanism.
Listen for misunderstanding as well as cancellation. Support questions such as “I thought this included all integrations” identify a practical failure that a plan-mix chart can hide.
Step 5 — Experiment with positioning
The E turns the framework into a decision. Specify the eligible audience, randomization unit, primary outcome, practical effect threshold, guardrails, and stopping rule before traffic enters the test. Keep assignment stable for returning buyers so they do not encounter contradictory offers.
Estimate sample size using the baseline and variability of the metric you actually need to improve. A test powered for plan clicks may be unable to resolve a commercially meaningful change in revenue or contribution. Choose a fixed-horizon analysis or an appropriate sequential method in advance.
Microsoft’s research on metric interpretation pitfalls explains why repeatedly checking an ordinary significance test and stopping at the first favorable result increases false positives. A universal “run for four weeks” instruction does not solve this problem.
Track completed purchases and realized value across all eligible randomized visitors. Plan mix among purchasers is a diagnostic measure. It excludes people who left, so it cannot establish the business outcome by itself. Safety monitoring continues throughout; clear billing errors or serious customer harm justify intervention without pretending an interrupted experiment produced a clean winner.
Read the revenue calculation correctly
Here is an illustrative calculation, not a client result. Suppose 150 new customers buy monthly plans: 120 choose Basic at $49 and 30 choose Pro at $149. That cohort produces $10,350 in initial monthly subscription revenue before refunds and costs.
If the same 150 customers instead split into 90 Basic and 60 Pro, the figure becomes $13,350, an increase of $3,000, or about 29%. This calculation assumes the total number of buyers stays constant.
That assumption is decisive. With the improved 60/40 mix, average monthly revenue per new customer is $89. If the treatment produces only 110 buyers, initial monthly revenue is $9,790, below the original $10,350. Premium share has doubled while cohort revenue has fallen.
Multiplying the two constant-customer scenarios by 18 gives $186,300 and $240,300. Those are hypothetical 18-month cohort totals assuming every customer keeps paying the same price. They are not monthly revenue, measured lifetime value, or profit. Churn, discounts, collections, service costs, and expansion would change the result.
The denominator and the observation window belong beside the headline. Without them, a pricing result is too easy to overstate.
FAQ
Does a third option always improve conversion?
No. It can clarify a tradeoff, create extra work, or expose poor differentiation. The number of plans alone does not predict the result. Test the actual offer with the audience expected to buy it.
What price gap should a decoy have?
There is no universal percentage. Strict asymmetric dominance requires one option to be no worse on the relevant dimensions and better on at least one. A cheaper, less capable plan usually introduces a genuine tradeoff. Evaluate the comparison honestly instead of forcing it into the decoy label.
What would make this test worth shipping?
The agreed business outcome clears its decision threshold, the measurement is trustworthy, and customer guardrails remain acceptable. An attractive premium share is insufficient when overall value is uncertain. Apply the five VALUE steps to one proposed comparison before expanding the experiment to additional tiers or audiences.
